1

Cigarette Smoking Increases Copy Number Alterations in Non-Small Cell Lung

Cancer

Yen-Tsung Huang1,2, Xihong Lin2, Yan Liu3, Lucian R. Chirieac4, Ray McGovern2, John

C. Wain5,6, Rebecca S. Heist5,7, Vidar Skaug8, Shanbeh Zienolddiny8, Aage Haugen8, Li

Su7, 9Edward A. Fox, Kwok-Kin Wong3 and David C. Christiani1,7,10

Affiliations: Departments of 1Epidemiology, 2Biostatistics and 7Environmental Health,

Harvard School of Public Health, 667 Huntington Avenue, Boston, MA 02115;

3Department of Medical Oncology and 9Molecular Diagnostics Laboratory, Dana-Farber

Cancer Institute, 44 Binney Street Boston, MA 02115; 4Department of Pathology,

Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115; 5Cancer Center,

6Thoracic Surgery Unit and 10Pulmonary and Critical Care Unit, Massachusetts General

Hospital, 55 Fruit Street, Boston, MA 02114; and 8Department of Biological and

Chemical Working Environment, National Institute of Occupational Health, P.O. Box

8149 Dep, N-0033 Oslo, Norway.

Corresponding author: David C. Christiani, MD, MPH, Department of Environmental

Health, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115; e-

mail: dchris@hsph.harvard.edu

Classification:

Major: Biological Sciences; Minor: Genetics

2

ABSTRACT

Cigarette smoking has been a well-established risk factor of lung cancer for decades.

How smoking contributes to tumorigenesis in lung remains not fully understood. Here we

report the results of a genome-wide study of DNA copy number and smoking pack-years

in a large collection of non-small cell lung cancer (NSCLC) tumors. Genome-wide

analyses of DNA copy number and pack-years of cigarette smoking were performed on

264 NSCLC tumors, which were divided into discovery and validation sets. The copy

numbers-smoking associations were investigated in three scales: whole-genome,

chromosome/arm and focal regions. We found that heavy cigarette smokers (>60 pack-

years) have significantly more copy number gains than non-/light smokers (≤60 pack-

years) (p=2.46×10-4), especially in 8q and 12q. Copy number losses tend to occur away

from genes in non-/light smokers (p=5.15×10-5) but not in heavy smokers (p=0.52). Focal

copy number analyses show that there are strong associations of copy number and

cigarette smoking pack-years in 12q23 (p=9.69×10-10) where IGF1 (insulin-like growth

factor 1) is located. All of the above analyses were tested in the discovery set and

confirmed in the validation set. DNA double-strand break assays using human bronchial

epithelial cell lines treated with cigarette smoke condensate (CSC) were also performed,

and indicated that CSC leads to genome instability in human bronchial epithelial cells.

We conclude that cigarette smoking leads to more copy number alterations, which may

be mediated by the genome instability.

3

\body INTRODUCTION

Lung cancer, of which 85% is non-small cell lung carcinoma (NSCLC), is the second

most common cancer and the leading cause of cancer-related death in the United States.(1)

The epidemiologic evidence supporting that cigarette smoking is an important factor in

causing lung cancer was reported almost six decades ago.(2-4) Moreover, lung cancer

mortality mirrors trends in tobacco use(5). Carcinogens derived from cigarette smoking

damage lung epithelium by oxidative stress and direct DNA damage.(6) Although there

has been progress in our understanding of lung carcinogenesis over the past few decades,

the knowledge of mechanisms by which cigarette smoking causes lung cancer remains

incomplete.

Profiles of copy number alterations (CNAs) in NSCLC have been studied.(7, 8)

However, what causes copy number (CN) changes remain unknown. Several mechanisms

of copy number changes have been proposed including homologous recombinations and

non-homologous mechanisms.(9, 10) Bacteria, yeast and human seem to share similar

mechanisms.(10) In bacteria, copy number alterations can be induced by environmental

stress to enable swifter evolution in response to such stress. In the cell population within

a tumor or precancerous lesion, similar stress such as hypoxia may induce copy number

change. Thus, it is plausible to hypothesize that cigarette smoking serves as an

environmental stress on the cells that leads to tumorigenesis by means of copy number

alterations.

Using the tumor cells separated from malignant pleural effusions, it was found

that gains of 11p were more frequent in smoking men than non-smoking men.(11)

Furthermore, another study identified a copy number-based genomic signature in resected

4

lung tumors for current smokers as compared to never smokers.(12) However, these

studies had significant limitations. First, discrete smoking status (smokers vs. non-

smokers) may not be an optimal indicator to capture the dose-response relationship

between cigarette smoking and copy number changes. Second, smoking may have

different implications on copy number depending on whether it induces gains or losses.

Third, the conclusions in the previous studies were drawn based on modest sample sizes.

Lastly, none of previous studies provide a biological explanation on how cigarette

smoking causes CNAs. To better investigate the relationship between cigarette smoking

and copy number alterations, we conducted a genome-wide study of copy numbers and

smoking pack-years in a large collection of resected NSCLC tumors. Our analyses cover

the association of cigarette smoking with copy numbers on three different scales: whole-

genome, chromosome/arm and focal copy numbers. The causal mechanism behind such

smoking-CNAs association was further explored in a human non-tumorigenic bronchial

cell line.

RESULTS

A total of 264 subjects were randomly divided into two data sets: discovery and

validation sets. The characteristics of the populations are similar (Table 1), indicating the

balance of the two sets. Two alternative data splittings were pursued to prevent from the

possibility that the results presented here are simply due to chance or to multiple

comparisons. (SI Appendix (Tables S1-S3)) To account for batch effects, we also

performed batch-adjusted analyses by normalization and explicitly adjusting for the batch

identity as a covariate in the regression. The batch effect-adjusted analyses showed

5

similar patterns to those without adjustment. (SI Appendix (Tables S4 and S5; Figure S1))

The analyses of smoking vs. copy number associations are outlined as three parts: on the

genome-wide scale, on the chromosome/arm specific scale and on the focal region scale.

Cigarette smoking and whole-genome copy number pattern

There is a significant increase in total events of copy number gains among heavy smokers

(>60 pack-years) (p=0.0080, 0.0095, and 2.5×10-4 for discovery, validation and both sets,

respectively), but no difference in copy number losses (Figure 1A and 1B). No significant

difference was observed in age, clinical stage, histology and gender between heavy and

light/non-smokers.

For copy number losses, G/T ratios in light/non-smokers (≤60 pack-years) are

significantly lower than the null ratio (i.e., the ratio when CNAs occur at random with

respect to the gene location) (p=0.011, 9.80×10-4, and 5.15×10-5 for discovery, validation

and both sets, respectively) but heavy smokers (>60 pack-years) show no difference

(p=0.78, 0.31 and 0.52, respectively) (Figure 1C and 1D). These results suggest that copy

number losses tend to occur away from genes but such tendency disappears in heavy

smokers. In contrast, there is no consistent pattern for copy number gains. Heavy smokers

seem to have more genes with copy number changes, especially in gains. (SI Appendix

(Figure S2))

Cigarette smoking and copy number pattern by chromosome/arm

The chromosome/arm-specific analyses suggest the majority responsible for the genome-

wide difference comes from chromosomes 8q (p =1.19×10-5 for total events of CN gains

between light and heavy smokers) and 12q (p =2.1×10-4) (Figure 2) as well as many

others (chromosomes 1, 3, 7, 10, 11, 16 and 17) (SI Appendix (Figure S3)). Similar

6

results were observed when genomic location was taken into account, especially in 8q

and 12q. (SI Appendix (Figure S4A)) The dose-response relationships between

continuous copy numbers and smoking pack-years are also significant in 8q (p=0.015)

and 12q (p=0.0025). (SI Appendix (Figure S4B)) These two regions are also found the

most signals in focal copy number analyses, as will be shown in the following.

Cigarette smoking and focal copy numbers

As stated in Materials and Methods, we performed single- and multiple-marker analyses

to investigate the association of cigarette smoking and focal copy numbers. In the moving

window 10-marker analyses, we selected the top 50 sets with smallest p values in the

discovery set and tested the 50 sets using the validation set (p<0.05). Using such criteria,

we identified one 10-marker set in 12q23 with p values of 9.69×10-10, which reached the

genome-wide significance. (Figure 3A) The region harbors a gene, IGF1 (insulin-like

growth factor 1) that plays an important role in tumorigenesis. (Figure 3B) In the single-

marker analyses, the most significant signals are also in the same region of 12q23: two

loci are in the intron between the last two exons of IGF1 and two loci are located

downstream of IGF1. (SI Appendix (Figure S5A and S5B, and Table S6)) The p value of

the 10-marker set identified in the 10-marker analyses and the corresponding p values and

R2 from the single-marker analyses were shown in Table 2. Compared with the single-

marker analyses, statistical power was gained from the 10-marker analyses by borrowing

information in the neighboring markers, accounting for correlation among the CNVs in

the marker set, reducing degrees of freedom of the test, and reducing the total number of

tests across the genome.

7

The dose-response relationship of copy number and smoking pack-years for the

four loci in 12q23.2 is shown in SI Appendix (Figure S5C-F), indicating a J-shape curve.

That is, beyond a certain threshold, the higher smoking pack-years, the more departure

from the neutral copy number. Notably, the threshold, about 60 pack-years, is consistent

with the cut-off used in the above analyses of whole-genome CNAs pattern.

In addition to 12q23, 3q24 and 8q24 are two additional regions that are potentially

associated with the pack-years of cigarette smoking from single-marker analyses. (SI

Appendix (Figures S6 and S7)) We also performed the analyses in the dichotomous

version, detail of which can be found in SI Appendix. (Table S7 and Figure S8)

DNA double-strand break assay

To investigate further the results of our statistical analyses, we determined whether

cigarette smoke could induce DNA double-stand breaks in cultured cells. To mimic

longer and heavier cigarette smoking conditions, we treated human nontumorigenic

bronchial epithelial cell HBEC 3KT with 0.04 and 0.4 μg/ml CSC for 24 hours. Under

these conditions, the survival rates are 96.9% and 95.7%, respectively, indicating the dose

of CSC and the length of treatment used in this study are not toxic to the cells (Figure

4A). To minimize background DNA double-strand breakage, we treated cells with CSC

right after the growth had reached confluence. Under these conditions, ~5% of non-CSC

treated cells still display double stand breaks (Figure 4C). When treated with 0.04 μg/ml

CSC for 24 hours, the percentage of cells with double strand breaks increased to 15%.

This percentage doubled with the application of more concentrated 0.4 μg/ml CSC

(Figure 4C). We also treated the cells with 0.4 μg/ml CSC for 2 hours, and observed a

8

similar DNA double-strand break ratio as that of the non-CSC treated control cells,

suggesting DNA double-strand break occurring after a longer time of CSC treatment.

To determine the effects of CSC on induction of cellular apoptosis which

indirectly contributes to DNA double-strand breaks, the same set of cells (as used in

Figure 4B and 4C) were lysed for apoptotic-specific Caspase-3/7 activity. As shown in

Figure 4D, there is a basal level of Caspase-3/7 activity in non-CSC treatment cells. Upon

CSC treatment, the value of relative fluorescence unit (RFU) increased in a dose-

dependent matter. However, the extent to which the RFU value increased in response to

CSC treatment was much less than the corresponding increase in DNA double strand

breaks in Figure 4C. Collectively, these results indicate that higher CSC leads to genome

instability in bronchial epithelial cells. As such, theses data provide biological evidence

to bridge the associations between CNAs and smoking observed in the above human data.

DISCUSSION

We show that heavy smokers (>60 pack-years) have more copy number gains than

light/non-smokers but not copy number losses and that light/non-smokers (≤60 pack-

years) have copy number losses away from the gene location, in contrast to heavy

smokers. The discrepancy between gains and losses suggests that different mechanisms

may exist for the genome impact of cigarette smoking. For gains, smoking executes its

oncogenic effect by increasing the event of copy number changes. For losses, in contrast,

smoking does not increase CNAs events but increase the proportion of genes being

affected. Because losing a fragment of DNA is less favorable than gaining one(13), two

separate mechanisms may be developed to hit the genes responsible for tumorigenesis.

9

The phenomenon may be a consequence of selection during cancer development and cell

proliferation. Because different cells possess different CNAs, selection by a nutrient-

limited environment makes those clones that can grow without regulatory control become

dominant.

For heavy smokers, there were more copy number gains compared to non- or light

smokers and no tendency for copy number losses to occur away from the gene location.

We have also found that genes with gains are more likely to be oncogenes or to be

involved in pathways that are associated with tumor growth, which suggests that lung

cancer cells in heavy smokers tend to acquire the growth advantage via copy number

gains.(14) As a result, copy number losses within genes have less unfavorable impact on

such cells since it is compensated by the fact that they can grow without regulation. This

explains our observation that the proportion of losses within gene among heavy smokers

is not different from that at random.

Previous studies have shown that copy number alterations are more frequent in

smokers than in non-smokers,(11, 12) consistent with our findings based on pack-years.

Copy number-based genomic signature has also been identified to discriminate current

smokers and never smokers(12), which, however, does not include IGF1. Smoking status

may not necessarily reflect the same oncogenic feature as pack-years of smoking, a

measure of cumulative exposure. Furthermore, the large sample size and discovery-

validation process in this study increase robustness of the findings.

Smoking causes lung cancer through numerous carcinogens derived from

cigarette combustion. There are two parts of the carcinogenic effect: early damage of

oxidative stress by reactive oxygen species and late damage by DNA adduct and DNA

10

mutation.(6) Both kinds of damage can serve as initiators of copy number changes,

especially oxidative stress. It has recently been proposed that cellular stress coming from

environmental agents can induce copy number changes, which seems to be a common

mechanism in bacteria, yeast and humans.(10)

The most significant region on 12q23 is at the junction of the last two exons and

the downstream of IGF1. The two loci within IGF1 are located in the intron between the

last two exons of IGF1. The protein product of the aberrant genomic DNA can exert its

undue influence on the cellular physiology. On the other hand, if the key player is the

downstream rather than the coding region of IGF1, it is still possible that IGF1 function

is affected because the downstream fragment can serve as a regulatory element of IGF1

transcription. That is, the alterations of the regulatory element can lead to the abnormal

gene expression of IGF1.

IGF1/IGF1R signaling pathway can induce many effects, including cell

proliferation, differentiation, transformation and inhibition of apoptosis.(15) Because of

the overlap with downstream signaling pathways of epidermal growth factor receptor

(EGFR) signaling, IGF1/IGF1R signaling may modulate the EGFR pathway, a critical

pathway in lung tumorigenesis(16), and it may explain, in part, clinical resistances to

EGFR inhibitors(17).

Several studies have provided the links among smoking, IGF1 and cancer. For

example, it has been reported that smoking may affect IGF1 serum level and its

signaling.(18, 19) On the other hand, IGF1 and the risk of developing cancer have also

been extensively studied in lung cancer(20-23), breast cancer(24), prostate cancer(25)

and colorectal cancer(26). Our analysis supports the hypothesis that smoking can act

11

through increasing the copy number of IGF1 to induce its over-expression and

subsequent oncogenesis.

MATERIALS AND METHODS

Study population, specimens and data collection

A series of 264 snap-frozen tumor samples from NSCLC patients with complete

information on cigarette smoking was collected during surgery or biopsy from the

Massachusetts General Hospital (MGH), Boston, MA, and the National Institute of

Occupational Health, Oslo, Norway. An additional 50 paired specimens of non-neoplastic

lung parenchyma and 63 paired blood samples were included as the reference group for

copy number estimation. Demographic and smoking information was collected by a

trained research assistant using a modified standardized American Thoracic Society

respiratory questionnaire.(27) A similar approach was used for the Norwegian cohort.(28)

Written informed consents were obtained from all patients. The study was approved by

the institutional review boards of MGH, the Harvard School of Public Health, and the

Norwegian Data Inspectorate, and Local Regional Committee for Medical Research

Ethics.

DNA quality, histopathology and genechip

DNA was extracted from tumor and non-neoplastic lung parenchyma after manual

microdissection from 5-μm thick histopathologic sections. Each specimen was evaluated

for amount and quality of tumor cells. Tumors were reviewed and classified using the

WHO criteria. Specimens with lower than 70% tumor cellularity, inadequate DNA

12

concentration, or not intact genomic DNA were not included for chip hybridization. The

platform of genechip is Affymetrix 250K Nsp GeneChip®.

Data preprocessing

Copy numbers were obtained with dChip software by invariant set normalization and

median smoothing with the window of 11 loci.(29) Only 256,554 probes on somatic

chromosomes were analyzed. We further classified the continuous inferred copy number

into a discrete variable of CNAs: copy number gains defined as copy numbers ≥2.7 and

copy number losses defined as copy numbers ≤1.3, to detect copy number ≥3 and ≤1 by

tolerating 30% normal tissue contamination. The probes were mapped to the RefSeq

genes with 2 kb extension both upstream and downstream using the UCSC Genome

Browser. Among the 256,554 probes on somatic chromosomes, 104,256 (40.64%) were

mapped to 11,700 genes.

Statistical analysis

Only early stage tumors were analyzed here because we have found that late stage tumors

have more CNAs. The number of pack-years is defined as the packs of cigarette smoked

per day multiplied by the years of smoking. 60 pack-years of cigarette smoking was

chosen as the cut-off for heavy and light/non-smokers according to the observation of

total CNAs events by the interval of 10 pack-years in both discovery and validation sets

(SI Appendix (Figure S9)). Using the cut-off, we had 203 light/non-smokers and 61

heavy smokers. We developed three methods to test the genome-wide or

chromosome/arm-specific copy number patterns between heavy and light/non-smokers

and one method to test the

association of the chromosome/arm-specific or focal CNs and smoking pack-years.

13

First, we calculated the total events of copy number gains and losses and

compared them between the two smoking groups by the two-sided t test, which provides

a convenient summary index but collapses CNAs information over the genomic locations.

The second method is to apply two-sample tests for continuous copy number by

calculating the standardized difference of two average copy numbers for each locus as:

( ) 221121 nvnvmmc iiiii +−= where mji and vji is the estimated mean and variance,

respectively, of copy number for group j at locus i, and nj is the sample size in group j.

We summed up ci 2 over i across the loci in the arm of a chromosome to calculate the

observed total standardized squared difference (Cobserved). By permuting the two groups

and carrying out the above procedure for 10,000 times, we obtained a non-parametric null

distribution (Cnull). Then p values were obtained by comparing Cobserved and Cnull. This

permutation procedure provides a valid global test for the overall difference by

accounting for multiple comparisons and correlation of CNAs between different loci.

The third is a similar method extended to the discrete variable of CNAs (CN≥2.7

or not; CN≤1.3 or not) as mentioned above, but with advantage of testing gains and losses

separately. We applied two-sample tests for binomial data by calculating the standardized

difference of two proportions for each locus as:

( ) ( ) ( ) 22211121 11 nppnppppd iiiiiii −+−−= where pji is the estimated proportion

(stabilized by adding 0.5 in the numerator) of CN gains (or losses) for group j at locus i

and nj is the sample size in group j. We summed up di 2 over i across the arm of a

chromosome to calculate the observed total standardized squared difference (Dobserved).

Non-parametric null distribution (Dnull) was approximated by 10,000 permutations, and p

values were obtained by comparing Dobserved and Dnull.

14

The above three methods require the smoking exposure to be dichotomized. To

fully capture the continuous dose-response relationship between copy numbers and pack-

years of cigarette smoking, we also developed a test to summarize such association in a

chromosome/arm-specific fashion. We obtained the test statistics, =i iobserved fF where

if is the F statistics of regressing continuous copy numbers on the smoking pack-years

(square-root transformed) up to the quadratic term at locus i. Again, the non-parametric

null distribution (Fnull) was generated by 10,000 permutations and p values were obtained

as the tail probability of Fobserved in Fnull.

The proposed test is equivalent to the powerful score test for testing the variance

of coefficients in a multivariate regression by assuming regression coefficients have an

arbitrary distribution with mean 0 and variance τ (30), in which copy numbers of a region

or chromosome (as a vector) are regressed on smoking pack-years. The null hypothesis of

our proposed test is that all the coefficients relating pack-years to copy numbers are zero,

or equivalently, copy numbers at all loci have no association with smoking pack-years,

which is equivalent to H0: τ=0. The alternative hypothesis would be that copy numbers at

some loci have association with smoking pack-years. This variance component test is a

powerful test by borrowing information in multiple markers and effectively accounting

for correlation among the CNVs in a marker set, and reducing the degrees of freedom of

the test.

We used another method to investigate gene selection of CNAs between heavy

and light/non-smokers. Both the total probes (T) in which CNAs were detected and the

probes locating within genes (G) in which CNAs were detected were calculated for each

individual. We proposed a ratio of G vs. T (termed as G/T ratio) to estimate the selection

15

of CNAs with respect to the gene location. Under the null hypothesis that CNAs occur

randomly relative to where genes locate, we would expect the null ratio of 40.64%

(104,256/256,554), where 104,256 is the number of probes located within genes on the

chip. By comparing the G/T ratios to the null ratio, 40.64%, with two-sided t test, we

were able to test whether CNAs occur preferentially away from genes.

To investigate the association of focal copy numbers and smoking pack-years, we

analyzed copy number >2 and ≤2 separately. Two outliers (>2.5 standard deviations) of

smoking pack-years were excluded to eliminate the potential spurious result driven by

them. Square root was taken for smoking pack-years to transform a right skewed

distribution into an approximately normal distribution.(31) Both multiple-marker and

single-marker analyses were performed. In the multiple-marker analyses, we grouped the

consecutive ten SNPs (markers) as a set and calculated Fobserved and Fnull by the methods

mentioned above to obtain the p values for each set of markers. For those with Fobserved

much greater than Fnull (from 10,000 permutations), the null distributions were obtained

by the Satterthwaite approximation(32), in which the first two moments of scaled χ2

distribution were matched with those of Fnull. In total, we performed 25,655 hypothesis

tests for copy number >2 and ≤2 separately. Such multiple-marker analyses had better

statistical power than the single-marker analyses when the markers were correlated,

which is the case in the copy number data.

For single-marker analyses, 256,554 regressions for both copy number >2 and ≤2

were performed in the discovery set with continuous copy number at each locus as a

dependent variable and square root of smoking pack-years and its quadratic term as

independent covariates. For the validated candidates (p<0.05 in the validation set), pooled

16

results were generated with linear regressions (with up to quadratic term of square root of

smoking pack-years), spline regressions (with spline of square root of smoking pack-

years) and locally weighted scatter plot smoothing (LOWESS). Adjusted linear and

spline regressions were performed with adjustment of age at diagnosis, gender, two

cohorts, clinical stage and histology.

DNA double-strand break assay

For the cytotoxicity analysis of cigarette smoke condensate, a human nontumorigenic

bronchial epithelial cell line HBEC 3KT was cultured in 12-well plates to confluence and

then treated with indicated concentration of cigarette smoke condensate (CSC) for 24

hours. Viable cells were monitored by MTT assay using the CellTiter 96 AQueous One

Solution Cell Proliferation Assay kit (Promega, Madison, WI). All assays were

performed in triplicate. For the neutral comet assay, HBEC 3KT was cultured in 100-mm

plates to confluence and then treated with indicated concentration of CSC for 24 hours.

Cells having DNA double-stand break were analyzed by Neutral comet assay using

CometAssay kit (TREVIGEN, Gaithersburg, MD). About 600~800 cells were viewed per

treatment. For the apoptosis analysis, HBEC 3KT was cultured in 100-mm plates to

confluence and then treated with indicated concentration of CSC for 24 hours. The status

of cellular apoptosis was determined using SendoLyteTM Homogeneous Rh110 Caspase-

3/7 Assay kit (ANASPEC, Fremont, CA). All apoptosis assays were performed in

triplicate.

17

ACKNOWLEDGEMENTS

We are indebted to the participants of Molecular and Genetic Analysis of Lung Cancer

Study; to Dr. Cheng Li of HSPH for his advice on data preprocessing; to the HSPH Lung

Cancer Study Group: Dr. Kofi Asomaning, Dr. Eugene Mark, Dr. Matthew Kulke, Dr.

Wei Zhou, Dr. Geoffrey Liu, Marcia Chertok, Andrea Shafer, Lauren Cassidy, Maureen

Convery, Salvatore Mucci; to Drs. Panos Fidias and Bruce A. Chabner and the physicians

and surgeons of the Massachusetts General Hospital Cancer Center; and to Dr. Lodve

Stangeland of Haukeland University Hospital, Norway for recruiting patients. This study

is supported by US National Institutes of Health Grants No. CA076404 and CA134294

(Y.-T. H. and X.L.), CA092824 (D.C.C.), CA074386 (D.C.C.) and CA090578 (D.C.C.),

and Norwegian Cancer Society (A.H.).

18

REFERENCES

1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, et al. (2009) Cancer statistics, 2009 CA Cancer J Clin 59, 225-249.

2. Bach PB (2009) Smoking as a factor in causing lung cancer JAMA 301, 539-541. 3. Wynder EL & Graham EA (1950) Tobacco smoking as a possible etiologic factor

in bronchiogenic carcinoma; a study of 684 proved cases J Am Med Assoc 143, 329-336.

4. Doll R & Hill AB (1954) The mortality of doctors in relation to their smoking habits; a preliminary report Br Med J 1, 1451-1455.

5. Dubey S & Powell CA (2009) Update in lung cancer 2008 Am J Respir Crit Care Med 179, 860-868.

6. Alavanja MC (2002) Biologic damage resulting from exposure to tobacco smoke and from radon: implication for preventive interventions Oncogene 21, 7365-7375.

7. Kim TM, Yim SH, Lee JS, Kwon MS, Ryu JW, et al. (2005) Genome-wide screening of genomic alterations and their clinicopathologic implications in non-small cell lung cancers Clin Cancer Res 11, 8235-8242.

8. Weir BA, Woo MS, Getz G, Perner S, Ding L, et al. (2007) Characterizing the cancer genome in lung adenocarcinoma Nature 450, 893-898.

9. van Gent DC & van der Burg M (2007) Non-homologous end-joining, a sticky affair Oncogene 26, 7731-7740.

10. Hastings PJ, Lupski JR, Rosenberg SM, & Ira G (2009) Mechanisms of change in gene copy number Nat Rev Genet 10, 551-564.

11. Yen CC, Liang SC, Jong YJ, Chen YJ, Lin CH, et al. (2007) Chromosomal aberrations of malignant pleural effusions of lung adenocarcinoma: different cytogenetic changes are correlated with genders and smoking habits Lung Cancer 57, 292-301.

12. Massion PP, Zou Y, Chen H, Jiang A, Coulson P, et al. (2008) Smoking-related genomic signatures in non-small cell lung cancer Am J Respir Crit Care Med 178, 1164-1172.

13. Emerson JJ, Cardoso-Moreira M, Borevitz JO, & Long M (2008) Natural selection shapes genome-wide patterns of copy-number polymorphism in Drosophila melanogaster Science 320, 1629-1631.

14. Huang Y-T, Lin X, Chirieac LR, McGovern R, Wain JC, et al. (2011) Impact of Disease Development, Genomic Location and Biological Function on Copy Number Alterations of Non-Small Cell Lung Cancer PLoS One (in press).

15. Furstenberger G & Senn HJ (2002) Insulin-like growth factors and cancer Lancet Oncol 3, 298-302.

16. Herbst RS, Heymach JV, & Lippman SM (2008) Lung cancer N Engl J Med 359, 1367-1380.

17. Morgillo F, Kim WY, Kim ES, Ciardiello F, Hong WK, et al. (2007) Implication of the insulin-like growth factor-IR pathway in the resistance of non-small cell lung cancer cells to treatment with gefitinib Clin Cancer Res 13, 2795-2803.

18. Kaklamani VG, Linos A, Kaklamani E, Markaki I, & Mantzoros C (1999) Age, sex, and smoking are predictors of circulating insulin-like growth factor 1 and insulin-like growth factor-binding protein 3 J Clin Oncol 17, 813-817.

19

19. Tannheimer SL, Ethier SP, Caldwell KK, & Burchiel SW (1998) Benzo[a]pyrene- and TCDD-induced alterations in tyrosine phosphorylation and insulin-like growth factor signaling pathways in the MCF-10A human mammary epithelial cell line Carcinogenesis 19, 1291-1297.

20. Yu H, Spitz MR, Mistry J, Gu J, Hong WK, et al. (1999) Plasma levels of insulin-like growth factor-I and lung cancer risk: a case-control analysis J Natl Cancer Inst 91, 151-156.

21. Wu X, Yu H, Amos CI, Hong WK, & Spitz MR (2000) Joint effect of insulin-like growth factors and mutagen sensitivity in lung cancer risk J Natl Cancer Inst 92, 737-743.

22. Lukanova A, Toniolo P, Akhmedkhanov A, Biessy C, Haley NJ, et al. (2001) A prospective study of insulin-like growth factor-I, IGF-binding proteins-1, -2 and -3 and lung cancer risk in women Int J Cancer 92, 888-892.

23. Spitz MR, Barnett MJ, Goodman GE, Thornquist MD, Wu X, et al. (2002) Serum insulin-like growth factor (IGF) and IGF-binding protein levels and risk of lung cancer: a case-control study nested in the beta-Carotene and Retinol Efficacy Trial Cohort Cancer Epidemiol Biomarkers Prev 11, 1413-1418.

24. Hankinson SE, Willett WC, Colditz GA, Hunter DJ, Michaud DS, et al. (1998) Circulating concentrations of insulin-like growth factor-I and risk of breast cancer Lancet 351, 1393-1396.

25. Chan JM, Stampfer MJ, Giovannucci E, Gann PH, Ma J, et al. (1998) Plasma insulin-like growth factor-I and prostate cancer risk: a prospective study Science 279, 563-566.

26. Ma J, Pollak MN, Giovannucci E, Chan JM, Tao Y, et al. (1999) Prospective study of colorectal cancer risk in men and plasma levels of insulin-like growth factor (IGF)-I and IGF-binding protein-3 J Natl Cancer Inst 91, 620-625.

27. Zhou W, Heist RS, Liu G, Asomaning K, Miller DP, et al. (2006) Second hand smoke exposure and survival in early-stage non-small-cell lung cancer patients Clin Cancer Res 12, 7187-7193.

28. Zienolddiny S, Campa D, Lind H, Ryberg D, Skaug V, et al. (2008) A comprehensive analysis of phase I and phase II metabolism gene polymorphisms and risk of non-small cell lung cancer in smokers Carcinogenesis 29, 1164-1169.

29. Zhao X, Li C, Paez JG, Chin K, Janne PA, et al. (2004) An integrated view of copy number and allelic alterations in the cancer genome using single nucleotide polymorphism arrays Cancer Res 64, 3060-3071.

30. Liu D, Lin X, & Ghosh D (2007) Semiparametric regression of multidimensional genetic pathway data: least-squares kernel machines and linear mixed models Biometrics 63, 1079-1088.

31. Zhou W, Liu G, Park S, Wang Z, Wain JC, et al. (2005) Gene-smoking interaction associations for the ERCC1 polymorphisms in the risk of lung cancer Cancer Epidemiol Biomarkers Prev 14, 491-496.

32. Satterthwaite FE (1946) An Approximate Distribution of Estimates of Variance Components Biometrics Bulletin 2, 110-114.

33. Kwee L, Liu D, Lin X, Ghosh D, and Epstein M. (2008) A powerful and flexible multilocus association test for quantitative traits Am J of Human Genetics, 82, 386-397.

20

FIGURE LEGENDS

Figure 1. Association of cigarette smoking and whole-genome copy numbers.

A and B, Among the 256,554 total probes, the proportion (%) with CNAs (A, gains; B,

losses) events by pack-years of cigarette smoking (NS/LS: non-/light smokers, HS: heavy

smokers). C and D, Mean and its 95% confidence interval of G/T ratios in the heavy

smokers (HS) and non-/light smokers (NS/LS) for copy number gains (C) and losses (D);

and the dashed lines represent the null G/T ratio on the chip (104,256/256,554=40.64%).

P values were used to test the indicated indices between HS and NS/LS with methods

described in Methods.

21

Figure 2. Association of cigarette smoking and chromosome/arm-specific copy numbers.

A, the p values are from analyses testing the association of copy number-gain events with

the smoking group (heavy vs. light-/non-smokers). B, the p values are from analyses

testing the association of copy numbers (>2) with pack-years of cigarette smoking. The

dashed line indicates p=0.05.

22

Figure 3. Association of cigarette smoking and 25,655 moving window10-marker focal

copy numbers.

A. A plot of -log10P of the association between smoking pack-years and 10-marker set

focal copy number, which is analyzed for copy number >2 (upper half) and ≤2 (lower

half), separately. B, P values of focal copy number analyses in 12q23. The black dots and

line indicate p values from 10-marker analyses, and the superimposed gray dots and line

indicate the corresponding ones from single-marker analyses.

23

Figure 4. Effects of cigarette smoke condensate (CSC) treatment on HBEC 3KT DNA

double-stand breaks and apoptosis. A, Cytotoxic effect of CSC on HBEC3KT survival.

HBEC3KT cells were cultured in 12 well plates to confluence and then incubated with 0,

0.04, 0.4, 0.8, 4, and 40 μg/mL CSC for 24 hours. Viable cells were monitored with MTT

assay. Live cells treated with 0 μg/mL CSC were defined as 100%. Percentage of live

cells verses CSC concentration was plotted. B and C, CSC treatment induces DNA

single/double-stand breaks in HBEC 3KT cells. HBEC 3KT cells were culture to

confluence in 100-mm plates and then treated with 0, 0.04, and 0.4 μg/mL CSC for 24

hours. Cells were harvested by trypsinization and DNA single/double-stand breaks were

analyzed by neutral comet assay. A representative photo with undamaged-DNA (bright

dot) and DNA with single/double-strand breaks (bright dot with an elongated tail) was

shown (B). About 600~800 cells per treatment were viewed, and percentage of DNA

single/double-strand breaks verses CSC dose was plotted (C). D, CSC treatment induces

apoptosis in HBEC 3KT cells. The same set of cells used in B and C was also analyzed

for caspase-3/7 activity. Columns are mean value of relative fluorescence unit (RFU). A

larger RFU value represents a higher caspase-3/7 activity and thus a stronger apoptotic

response. Standard deviations are provided in C and D.

24

TABLE LEGENDS

Table 1. Characteristics of study populations

Table 2. Summary of the ten candidate loci at 12q23 from both 10-marker and single-

marker analyses.

1

Table 1. Characteristics of study populations

Discovery set Validation set P value*

Sample size 134 130

Male (%) 65.67 56.92 0.18

Age,

Mean ± standard deviation 67.27 ± 8.17 67.59 ± 8.39 0.75

Cigarette smoking pack-years,

Median ± interquartile range 34.25 ± 39.64 38 ± 35.93 0.28

Clinical stage 0.43

Stage 1 (%) 77.27 70.00

Stage 2 (%) 15.15 19.23

Stage 3 or 4 (%) 7.58 10.77

Cigarette smoking status 0.23

Never smokers (%) 7.46 6.15

Ex-smokers (%) 43.28 53.85

Current smokers (%) 49.25 40.00

Adenocarcinoma (%) 67.91 64.62 0.66

*P values were calculated with X2 test for percentage of male (1 degree of freedom, d.f.), adenocarcinoma,

patentis from MGH (1 d.f.), clinical stage (2 d.f.) and cigarette smoking status (2 d.f.); with t test for age;

and with Wilcoxon test for cigarette smoking pack-years.

2

1

Table 2. Summary of the ten candidate loci at 12q23 from both 10-marker and single-marker analyses.

Affy ID dbSNP

Cyto-

band

Position

(Mb) Gene

Focal copy number-smoking association

10-marker analyses Single-marker analyses

p value,

discovery set

p value,

validation set

p value,

pooled

p value,

pooled R2

p value,

adjusted*

SNP_A-2002985 rs5011687 12q23 101.157 -

3.17×10-8 0.0291 9.69×10-10

0.0152 0.065 0.0168

SNP_A-2125858 rs17439974 12q23 101.171 - 0.0239 0.060 0.0285

SNP_A-4222341 rs17032384 12q23 101.179 - 0.000110 0.126 0.000263

SNP_A-1899321 rs1520223 12q23 101.229 - 2.92×10-6 0.175 8.07×10-6

SNP_A-4222344 rs4764695 12q23 101.260 - 4.55×10-6 0.167 1.33×10-5

SNP_A-4228436 rs10860860 12q23 101.283 - 1.79×10-8 0.223 9.78×10-8

SNP_A-2106083 rs2946831 12q23 101.289 - 1.29×10-8 0.235 2.63×10-8

SNP_A-2255731 rs10745940 12q23 101.300 IGF1 3.26×10-6 0.163 8.20×10-6

SNP_A-2092658 rs9308315 12q23 101.306 IGF1 2.10×10-7 0.202 7.74×10-7

SNP_A-2271065 rs2072592 12q23 101.316 IGF1 6.17×10-7 0.200 4.42×10-6

*P values of smoking pack-years were calculated from linear models with up to quadratic term of square root-transformed smoking pack-years, adjusting for age,

gender, clinical stage, and cell type.

A B

C D

A

B

A

B

A B

C D